def test_setitem2():
d = Distribution(['00', '11'], [1, 0])
d.make_sparse()
d['11'] = 1/2
d.normalize()
assert_true('11' in d)
assert_almost_equal(d['11'], 1/3)
def test_marginalize():
outcomes = ['000', '011', '101', '110']
pmf = [1/4]*4
d = Distribution(outcomes, pmf)
d1 = d.marginal([0, 2])
d2 = d.marginalize([1])
assert_true(d1.is_approx_equal(d2))
def test_init1():
# Invalid initializations.
with pytest.raises(InvalidDistribution):
Distribution([])
with pytest.raises(InvalidDistribution):
Distribution([], [])
Distribution([], [], sample_space=[(0, 1)], validate=False)
def test_outcome_length():
outcomes = ['000', '011', '101', '110']
pmf = [1/4]*4
d = Distribution(outcomes, pmf)
d = d.marginal([0, 2])
assert_equal(d.outcome_length(), 2)
assert_equal(d.outcome_length(masked=True), 3)
def test_outcome_length():
outcomes = ['000', '011', '101', '110']
pmf = [1 / 4] * 4
d = Distribution(outcomes, pmf)
d = d.marginal([0, 2])
assert d.outcome_length() == 2
assert d.outcome_length(masked=True) == 3
def test_setitem2():
d = Distribution(['00', '11'], [1, 0])
d.make_sparse()
d['11'] = 1/2
d.normalize()
assert '11' in d
assert d['11'] == pytest.approx(1/3)
def test_marginalize():
outcomes = ['000', '011', '101', '110']
pmf = [1 / 4] * 4
d = Distribution(outcomes, pmf)
d1 = d.marginal([0, 2])
d2 = d.marginalize([1])
assert d1.is_approx_equal(d2)
def test_init8():
outcomes = [(0,), (1,)]
pmf = [1 / 2] * 2
d1 = ScalarDistribution(pmf)
d2 = Distribution.from_distribution(d1)
d3 = Distribution(outcomes, pmf)
assert d2.is_approx_equal(d3)
def test_init9():
outcomes = [(0,), (1,)]
pmf = [1 / 2] * 2
d1 = ScalarDistribution(pmf)
d2 = Distribution.from_distribution(d1, base=10)
d3 = Distribution(outcomes, pmf)
d3.set_base(10)
assert d2.is_approx_equal(d3)
def test_copy():
outcomes = ['0', '1']
pmf = [1/2, 1/2]
d1 = Distribution(outcomes, pmf)
d2 = d1.copy(base=10)
d3 = Distribution(outcomes, pmf)
d3.set_base(10)
assert_true(d2.is_approx_equal(d3))
def test_init9():
outcomes = [(0,), (1,)]
pmf = [1/2, 1/2]
d1 = ScalarDistribution(pmf)
d2 = Distribution.from_distribution(d1, base=10)
d3 = Distribution(outcomes, pmf)
d3.set_base(10)
assert_true(d2.is_approx_equal(d3))
def test_atoms():
pmf = [.125, .125, .125, .125, .25, 0, .25]
outcomes = ['000', '011', '101', '110', '222', '321', '333']
d = Distribution(outcomes, pmf)
atoms = outcomes
assert_equal( list(d.atoms()), atoms)
patoms = ['000', '011', '101', '110', '222', '333']
assert_equal( list(d.atoms(patoms=True)), patoms)
d = Distribution(outcomes, pmf, sample_space=outcomes + ['444'])
atoms = outcomes + ['444']
assert_equal( list(d.atoms()), atoms)
def test_init8():
outcomes = [(0,), (1,)]
pmf = [1/2, 1/2]
d1 = ScalarDistribution(pmf)
d2 = Distribution.from_distribution(d1)
d3 = Distribution(outcomes, pmf)
assert d2.is_approx_equal(d3)
def test_init2():
# Cannot initialize with an iterator.
# Must pass in a sequence for outcomes.
outcomes = map(int, ['0', '1', '2', '3', '4'])
pmf = [1 / 5] * 5
with pytest.raises(TypeError):
Distribution(outcomes, pmf)
def test_copy():
outcomes = ['0', '1']
pmf = [1 / 2] * 2
d1 = Distribution(outcomes, pmf)
d2 = d1.copy(base=10)
d3 = Distribution(outcomes, pmf)
d3.set_base(10)
assert d2.is_approx_equal(d3)
def test_atoms():
pmf = [0.125, 0.125, 0.125, 0.125, 0.25, 0, 0.25]
outcomes = ['000', '011', '101', '110', '222', '321', '333']
d = Distribution(outcomes, pmf)
atoms = d._product(['0', '1', '2', '3'], repeat=3)
assert list(d.atoms()) == list(atoms)
patoms = ['000', '011', '101', '110', '222', '333']
assert list(d.atoms(patoms=True)) == patoms
ss = CartesianProduct.from_outcomes(outcomes + ['444'])
d = Distribution(outcomes, pmf, sample_space=ss)
atoms = d._product(['0', '1', '2', '3', '4'], repeat=3)
assert list(d.atoms()) == list(atoms)
def test_setitem2():
d = Distribution(['00', '11'], [1, 0])
d.make_sparse()
d['11'] = 1 / 2
d.normalize()
assert '11' in d
assert d['11'] == pytest.approx(1 / 3)
def test_setitem2():
d = Distribution(['00', '11'], [1, 0])
d.make_sparse()
d['11'] = 1 / 2
d.normalize()
assert_true('11' in d)
assert_almost_equal(d['11'], 1 / 3)
def test_zipped():
pmf = [.125, .125, .125, .125, .25, 0, .25]
outcomes = ['000', '011', '101', '110', '222', '321', '333']
d = Distribution(outcomes, pmf)
outcomes_, pmf_ = list(zip(*d.zipped()))
d2 = Distribution(outcomes_, pmf_)
assert_true(d.is_approx_equal(d2))
outcomes_, pmf_ = list(zip(*d.zipped(mode='atoms')))
d3 = Distribution(outcomes_, pmf_)
assert_true(d.is_approx_equal(d3))
outcomes_, pmf_ = list(zip(*d.zipped(mode='patoms')))
d4 = Distribution(outcomes_, pmf_)
d.make_sparse()
np.testing.assert_allclose(d.pmf, d4.pmf)
def test_atoms():
pmf = [.125, .125, .125, .125, .25, 0, .25]
outcomes = ['000', '011', '101', '110', '222', '321', '333']
d = Distribution(outcomes, pmf)
atoms = d._product(['0','1','2', '3'], repeat=3)
assert_equal(list(d.atoms()), list(atoms))
patoms = ['000', '011', '101', '110', '222', '333']
assert_equal(list(d.atoms(patoms=True)), patoms)
ss = CartesianProduct.from_outcomes(outcomes + ['444'])
d = Distribution(outcomes, pmf, sample_space=ss)
atoms = d._product(['0','1','2', '3', '4'], repeat=3)
assert_equal(list(d.atoms()), list(atoms))
def test_init10():
d1 = Distribution([(0, 0), (0, 1), (1, 0)], [0.5, 0.25, 0.25])
pmf = [[0.5, 0.25], [0.25, 0]]
d2 = Distribution.from_ndarray(pmf)
assert d1.is_approx_equal(d2)
def test_set_rv_names2():
outcomes = ['00', '11']
pmf = [1 / 2, 1 / 2]
d = Distribution(outcomes, pmf)
assert_raises(ditException, d.set_rv_names, 'XYZ')
def test_init6():
outcomes = {'0', '1', '2'}
pmf = [1 / 3] * 3
with pytest.raises(ditException):
Distribution(outcomes, pmf)
def test_init4():
dist = {'0': 1 / 2, '1': 1 / 2}
pmf = [1 / 2] * 2
with pytest.raises(InvalidDistribution):
Distribution(dist, pmf)
def test_from_rv():
rv = sps.rv_discrete(values=([1, 2], [3 / 10, 7 / 10]))
d1 = Distribution.from_rv_discrete(rv)
d2 = Distribution([(1,), (2,)], [3 / 10, 7 / 10])
assert d1.is_approx_equal(d2)
def test_set_rv_names2():
outcomes = ['00', '11']
pmf = [1 / 2] * 2
d = Distribution(outcomes, pmf)
with pytest.raises(ditException):
d.set_rv_names('XYZ')
def test_is_homogeneous2():
outcomes = ['00', '01']
pmf = [1 / 2] * 2
d = Distribution(outcomes, pmf)
assert not d.is_homogeneous()
def test_is_homogeneous1():
outcomes = ['00', '11']
pmf = [1 / 2, 1 / 2]
d = Distribution(outcomes, pmf)
assert d.is_homogeneous()
def test_coalesce():
outcomes = ['000', '011', '101', '110']
pmf = [1/4]*4
d = Distribution(outcomes, pmf)
d = d.coalesce([[0, 1], [2]])
assert_equal(d.outcome_length(), 2)
def test_has_outcome1():
d = Distribution(['0', '1'], [1, 0])
d.make_sparse()
assert_false(d.has_outcome('1', null=False))
def test_init7():
outcomes = ['0', '1']
pmf = [1/2, 1/2]
d1 = Distribution(outcomes, pmf)
d2 = Distribution.from_distribution(d1)
assert_true(d1.is_approx_equal(d2))
def test_has_outcome2():
d = Distribution(['0', '1'], [1, 0])
assert_false(d.has_outcome('1', null=False))
def test_is_homogeneous2():
outcomes = ['00', '01']
pmf = [1/2, 1/2]
d = Distribution(outcomes, pmf)
assert_false(d.is_homogeneous())
def test_zipped():
pmf = [0.125, 0.125, 0.125, 0.125, 0.25, 0, 0.25]
outcomes = ['000', '011', '101', '110', '222', '321', '333']
d = Distribution(outcomes, pmf)
outcomes_, pmf_ = list(zip(*d.zipped()))
d2 = Distribution(outcomes_, pmf_)
assert d.is_approx_equal(d2)
outcomes_, pmf_ = list(zip(*d.zipped(mode='atoms')))
d3 = Distribution(outcomes_, pmf_)
assert d.is_approx_equal(d3)
outcomes_, pmf_ = list(zip(*d.zipped(mode='patoms')))
d4 = Distribution(outcomes_, pmf_)
d.make_sparse()
assert np.allclose(d.pmf, d4.pmf)
def test_has_outcome2():
d = Distribution(['0', '1'], [1, 0])
assert not d.has_outcome('1', null=False)
def test_setitem1():
d = Distribution(['0', '1'], [1 / 2] * 2)
with pytest.raises(InvalidOutcome):
d.__setitem__('2', 0)
def test_init3():
dist = {'0': 1 / 2, '1': 1 / 2}
d = Distribution(dist)
assert_equal(d.outcomes, ('0', '1'))
assert_array_almost_equal(d.pmf, [1 / 2, 1 / 2])
def test_init10():
d1 = Distribution([(0, 0), (0, 1), (1, 0)], [0.5, 0.25, 0.25])
pmf = [[0.5, 0.25], [0.25, 0]]
d2 = Distribution.from_ndarray(pmf)
assert d1.is_approx_equal(d2)
def test_init1():
# Invalid initializations.
assert_raises(InvalidDistribution, Distribution, [])
assert_raises(InvalidDistribution, Distribution, [], [])
Distribution([], [], sample_space=[(0, 1)], validate=False)
def test_from_rv():
rv = sps.rv_discrete(values=([1, 2], [3/10, 7/10]))
d1 = Distribution.from_rv_discrete(rv)
d2 = Distribution([(1,), (2,)], [3/10, 7/10])
assert d1.is_approx_equal(d2)
def test_init3():
dist = {'0': 1 / 2, '1': 1 / 2}
d = Distribution(dist)
assert d.outcomes == ('0', '1')
assert np.allclose(d.pmf, [1 / 2] * 2)
def test_coalesce():
outcomes = ['000', '011', '101', '110']
pmf = [1 / 4] * 4
d = Distribution(outcomes, pmf)
d = d.coalesce([[0, 1], [2]])
assert d.outcome_length() == 2
def test_init5():
outcomes = ['0', '1', '2']
pmf = [1 / 2] * 2
with pytest.raises(InvalidDistribution):
Distribution(outcomes, pmf)
def test_set_rv_names2():
outcomes = ['00', '11']
pmf = [1/2, 1/2]
d = Distribution(outcomes, pmf)
with pytest.raises(ditException):
d.set_rv_names('XYZ')
def test_init7():
outcomes = ['0', '1']
pmf = [1 / 2] * 2
d1 = Distribution(outcomes, pmf)
d2 = Distribution.from_distribution(d1)
assert d1.is_approx_equal(d2)
def test_setitem1():
d = Distribution(['0', '1'], [1/2, 1/2])
with pytest.raises(InvalidOutcome):
d.__setitem__('2', 0)
def test_is_homogeneous2():
outcomes = ['00', '01']
pmf = [1 / 2, 1 / 2]
d = Distribution(outcomes, pmf)
assert_false(d.is_homogeneous())
def test_has_outcome1():
d = Distribution(['0', '1'], [1, 0])
d.make_sparse()
assert not d.has_outcome('1', null=False)
def test_setitem1():
d = Distribution(['0', '1'], [1 / 2, 1 / 2])
assert_raises(InvalidOutcome, d.__setitem__, '2', 0)
def test_is_homogeneous1():
outcomes = ['00', '11']
pmf = [1/2, 1/2]
d = Distribution(outcomes, pmf)
assert_true(d.is_homogeneous())